Tentative steps in mining UK 
theses
OR 2016, Dublin
June 2016
www.bl.uk 2
Is there valuable content in theses?
“Anything worthwhile in a thesis would 
have been published separately anyway.”
-- bioscience researcher
www.bl.uk 3
UK PhD theses
• Cutting edge research
• Not published elsewhere
• Traditionally book, now usually e-
• PDF – but new forms emerging
• 20,000 / year
• 300 pages each
• 6m pages of unique research every 
year
www.bl.uk 4
EThOS – e-theses online service
www.bl.uk 5
www.bl.uk 6
UK thesis collection & EThOS
http://ethos.bl.uk
www.bl.uk 7
Theses by Date
1% 12%
33%
54%
Pre-20th Century
1900-1949
1950-1979
1980-1999
2000-2016
www.bl.uk 8
Theses by Subject
0
10000
20000
30000
40000
50000
60000
70000
www.bl.uk 9
TDM examples
www.bl.uk 10
Alzheimer’s Society report
http://
www.rand.org/randeurope/research/projects/mapping-uk-dem
entia-research-landscape.html
 
www.bl.uk 11
TDM case study - Alzheimer’s Society & 
RAND Europe
Mapping the UK’s Dementia 
Research Landscape
- Workforce pipeline
- Tracked PhD to senior 
research
- 1/5 dementia PhD graduates 
remain in dementia research
- 70% leave dementia research 
within 4 years of completing 
PhD
- Used EThOS metadata to 
analyse trends
http://britishlibrary.typepad.co.uk/science/2015/09/a-novel-use-of-phd-data.html
www.bl.uk 12
Dementia search terms
• Alzheimer’s 
• Dementia
• Cognitive impairment
• Mixed dementia 
• Early onset dementia
• Vascular dementia
• Lewy bodies (Dementia with Lewy bodies)
• Frontotemporal dementia
• Posterior Cortical Atrophy
• Familial dementia
• Creutzfeldt Jakob
• Korsakoff’s syndrome
• Cognitive impairment
• Supranuclear palsy
• Binswanger’s
• Multiple sclerosis
• Motor neurone disease
• Parkinson’s
• Huntington’s
www.bl.uk 13
FLAX Interactive Language Learning
• http://flax.nzdl.org/greenstone3/flax?a=fp&sa=library
• Article - 
http://www.journals.elsevier.com/learning-culture-and-social-interaction
/
 
www.bl.uk 14
TDM case study – FLAX interactive 
language learning
• Model writing at research level; domain-specific texts; co-
located phrases
• Auto extraction & re-use for language learning
• Used EThOS metadata abstracts
• University of Waikato & Queen Mary, London
www.bl.uk 15
Metadata or full text theses?
Metadata Full texts
Content 400,000 records 130,000 theses
Format
Data - Digitised from print
- E-born
File format Xml or Excel PDF, .wav, .mov …
Access
- Harvest via OAI-PMH
- Supplied data
- Download from EThOS or other 
repository
- Supplied with permissions
Rights In the public domain Rights holders
www.bl.uk 16
TDM case study – National Compound 
Collection
• Are there useful molecules in PhD 
theses?
• Extract the compounds; re-draw in 
ChemDraw; input into ChemSpider
• Bristol Uni & Royal Society Chemistry
• Manual pilot – could process be 
automated?
• Used theses “likely to reveal new 
compounds”
• 47k compounds discovered (50% new)
www.bl.uk 17
Data collection
N-(3,5-Dinitrophenyl)-2-[(5-methyl-3,4-diphenyl-
1H-pyrrol-2-yl)carbonyl]hydrazinecarboxamide
Louise Sarah Evans, University of Southampton, 2006
Data 
Collectors
Theses
Molecular Structures
Open Access Database
> 45,000 
compounds
www.bl.uk 19
EThOS – http://ethos.bl.uk 
• Metadata for all UK doctoral (PhD) theses
• 430,000 records
• Top quality, accurate, consistent, unduplicated metadata
• Unique research, often not published elsewhere, cutting edge
• Data includes:
– Author, title, year, university name
– Abstracts (for 40%)
– Supervisor names, funder/sponsor body
– A few DOI and ORCiD identifiers
– Subject discipline.
www.bl.uk 20
Summary - EThOS data available
• Excel or XML via OAI-PMH harvest:
http://simba.cs.uct.ac.za/~
ethos/cgi-bin/OAI-XMLFile-2.21/XMLFile/ethos/oai.pl 
• Data.bl.uk (coming soon)
www.bl.uk 21
Thank you
Sara.Gould@bl.uk
ethos.bl.uk